Voice and music · by Google DeepMind

Gemma

Run an offline assistant that also hears speech and reads images.

text, image, video → text, data, code · speech → text · image, document → text, data
Way in
Official APINo sign-up, no key
To start
Freeopen weights (Apache-2.0)
Runs
On your MacAlso through Ollama, lmstudio, mlx-community, Vertex AI, OpenRouter
Checked
Nov 3, 2026Prices as of that day

What it can do 3

Chat, reason and call tools locally (Gemma 4 12B / E4B)gemma4:12b-it-qat 7.2 GB (q4_K_M 8.0 GB); e4b-it-qat 6.1 GB; e2b-it-qat 4.3 GB; 26B-A4B (16 GB) and 31B do not fittext, image, video → text, data, codefree (local)
Transcribe or translate speech (audio input)speech → textfree (local)
Understand images, charts, documentsimage, document → text, datafree (local)

Does the same job

Prices and limits as of Nov 3, 2026, from Google DeepMind’s own pages. They change; check before you rely on them.Not affiliated with Google DeepMind. Not advice.Something wrong?
Made by Jasper Burns

A short note when I ship something new, about once a month. One list for all my projects.